Table 1. Univariate regressions explaining the variation in species richness of all Macaronesian bryophytes (STOT), mosses (SM) and liverworts (SL) as a function of the predictors chosen for the Equilibrium Model of Island Biogeography (EMIB), the General Dynamic Model (GDM), the Habitat Diversity model (HD) and the Climatic Model (CLIMATE).
All bryophytes (STOT) | Mosses (SM) | Liverworts (SL) | ||||
R2 | F | R2 | F | R2 | F | |
EMIB | ||||||
A (+) | 0.132 | 2.59 | 0.222 | 4.85* | 0.015 | 0.25 |
DM | 0.002 | 0.43 | 0.009 | 0.16 | 0.102 | 1.94 |
DI | 0.017 | 0.30 | 0.037 | 0.65 | <0.001 | 0.00 |
N | 0.079 | 1.46 | 0.064 | 1.16 | 0.087 | 1.62 |
GDM | ||||||
T | 0.094 | 1.76 | 0.047 | 0.85 | 0.189 | 3.97† |
TT2 | 0.191 | 1.88 | 0.154 | 1.46 | 0.243 | 2.57 |
HD | ||||||
ELEV (+) | 0.405 | 11.58** | 0.482 | 15.83*** | 0.201 | 4.26* |
sdELEV (+) | 0.505 | 17.32*** | 0.606 | 26.13*** | 0.242 | 5.42* |
SLOPEdiv | 0.045 | 0.79 | 0.050 | 0.89 | 0.023 | 0.39 |
EZ (+) | 0.644 | 39.75*** | 0.579 | 23.39*** | 0.224 | 4.92* |
CLIMATE | ||||||
TMAX (−) | 0.195 | 4.12† | 0.082 | 1.53 | 0.467 | 14.87** |
TS (+) | 0.275 | 6.45* | 0.234 | 5.19* | 0.294 | 7.10* |
PMIN | 0.004 | 0.07 | 0.005 | 0.09 | 0.103 | 1.96 |
PS | 0.006 | 0.10 | 0.006 | 0.10 | 0.130 | 2.53 |
PANN (+) | 0.033 | 0.58 | 0.002 | 0.03 | 0.187 | 3.90† |
MistL (+) | 0.584 | 23.90*** | 0.503 | 17.20*** | 0.610 | 26.62*** |
p<0.06,
p<0.05,
p<0.01,
p<0.001.
Variable codes: A (area), DM (distance to mainland), DI (distance to the nearest island), N (neighbour index); T (oldest geological age); ELEV (maximum elevation), sdELEV (standard deviation of elevation), SLOPEdiv (diversity of slopes); EZ (number of ecological zones); TMAX (maximum temperature of warmest month), TS (temperature seasonality), PMIN (precipitation of driest quarter), PS (precipitation seasonality), PANN (annual precipitation), MistL (orographic mist layer).
The explanatory capacity of each variable (R2) and its statistical significance (F-test) are shown. The sign of the relationship is indicated under parenthesis after the predictor variable only when there is a significant effect on the dependent variable. The best fitting function (including significant quadratic functions) is shown in all cases, except for GDM for which both linear (T) and quadratic (TT2) functions of time are included as suggested in the literature [8], [10].